POT NORMALIZED VARIANCE PARAMETER SEARCH OF THE TEMPORAL NEYMAN-SCOTT RAINFALL MODEL

Abstract

In previous investigations on the Neyman-Scott clustered Poisson rectangular pulse rainfall model (NSM), model parameters were often estimated using rainfall moments, covariances, correlations, and dry moments while model performance was often assessed through the accuracy of extreme values. No former contribution included extreme value information in improving the parameter search. This was the motivation of the authors in finding a link between the scaling apparent in the historical rainfall maxima and the NSM parameters. The normalized variance or Fano factor of the Peaks Over Threshold (POT) rainfall process was adopted to define this link. In general, the results obtained from this approach showed acceptable performance with minor limitations.